Date:         Thu, 25 Jul 2002 14:02:16 -0700
Reply-To:     Warren <wmay@PREVMED.UMSMED.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Warren <wmay@PREVMED.UMSMED.EDU>
Organization: http://groups.google.com/
Subject:      Re: matched pair sample in SAS
Content-Type: text/plain; charset=ISO-8859-1

Theodore,
One method of matching that makes some sense is to use propensity
scores.  Basically, you run a logistic regression with your two
"groups" as the dependent variable and your age, education, etc.
covariates.  There are various articles that discuss the method, so do
a google search on something like...propensity matched pairs.  There
are some algorithms out there, also, that use a little different
approach and weight the covariates so that you can get good matches
on, say, gender with a little latitude on the age variable.

Luck
Warren

"Theodore Walls" <theodore.walls@yale.edu> wrote in message news:<169981dc9a711102baec5e490ec6f4ff.104135@mygate.mailgate.org>...
> I have two independent samples.  Instead of an independent samples
> ttest, I want to run a matched pair ttest in sas using a number of
> covariates to set up the pairs, such as age, education, etc.  Does
> anyone know how to set up the pairs in the data step?